 |
Giant leap in small technology:
Milestone in the field of nanoelectronics
By David Pescovitz
In the last decade, startling advances in nanoscience
promised to open vast new horizons for the future of computing.
The ability to understand and control matter at the atomic scale
could someday lead to powerful new devices, including handheld
sensors for sniffing out the tiniest traces of pathogens and massively
dense computer memory chips that outshine today’s state-of-the-art
devices by an order of magnitude. Making these wee wonders a reality,
though, requires bridging the nanoworld with the microworld of
traditional integrated circuit technology.
|
Professor
Jeffrey Bokor stands in front of a poster depicting the layout
of the RANT chip's carbon nanotubes grown on top of the silicon
circuit.
PEG SKORPINSKI PHOTO |
In January, Berkeley and Stanford University researchers announced
this giant leap in small technology with their fabrication of
the world’s first integrated circuit combining carbon nanotube
transistors and silicon transistors on the same chip.
“This is a critical first step in building the most advanced
nanoelectronic products, in which we would put carbon nanotubes
on top of a powerful silicon integrated circuit so they can interface
with an underlying information processing system,” says
Jeffrey Bokor, Berkeley professor of electrical engineering and
computer sciences and principal investigator of the project.
Carbon nanotubes are carbon molecules that resemble rolls of chicken
wire except that they are little more than a nanometer—just
a billionth of a meter—in diameter. Depending on how they’re
grown, carbon nanotubes can either be semiconducting or metallic,
making them well suited for electronic applications like nanowires
and nanotransistors pioneered in the last few years by Bokor’s
collaborator Hongjie Dai, a professor of chemistry at Stanford
University, and independently by researchers with Berkeley’s
physics department, the Lawrence Berkeley National Laboratory,
and elsewhere. Because of their diminutiveness, carbon nanotube
transistors can be packed far more closely together on a chip
than their silicon counterparts, providing much more processing
power or memory in the same amount of space.
An ongoing challenge is creating just the right chemistry to selectively
grow defect-free carbon nanotubes with the desired electrical
properties and control their placement. Researchers were unable
to predict even the proportion of metallic and semiconducting
nanotubes grown in each batch. As a result, each individual carbon
nanotube had to be electronically probed by hand one at a time
to determine its electrical properties. Quite simply, growing
carbon nanotubes required far too much trial and error for practical
electronic applications—until now.
The novel hybrid chip fabricated by Bokor, Dai, Berkeley graduate
student Yu-Chih Tseng, Stanford graduate student Ali Javey, and
their collaborators automates that painstaking process.
“We succeeded in making a tool for nanotechnology researchers,
and in the process we demonstrated the broader proof of principle
that nanotubes can be successfully integrated in a complex circuit,”
Tseng says.
 |
After
it is fabricated, the RANT chip is connected to the probing
pins of an automatic chip testing machine, known as an autoprobe.
PEG SKORPINSKI PHOTO |
The random access nanotube test chip, or RANT chip, was fabricated
in a two-part process that began in Berkeley’s Microfabrication
Lab. Using traditional integrated circuit patterning techniques,
approximately 4,000 transistors were etched into a silicon wafer.
After the transistors were patterned, wires had to be added that
connect the transistors with each other, and later, the carbon
nanotubes.
“Our challenge was to build an interconnect that would
work for both silicon and the nanotubes,” Tseng says.
The carbon nanotubes are grown directly onto “islands”
containing the specific catalyst necessary for nanotube synthesis.
It’s a very hot process though, taking place in a one-inch
furnace that climbs to temperatures of 875 degrees Celsius. Standard
interconnect materials like aluminum and copper melt at such high
temperatures. To prevent the circuitry from burning up, the researchers
used molybdenum, a refractory metal that can withstand the heat
of the furnace.
 |
Graduate
student Yu-Chih Tseng uses the autoprobe to test the RANT
chip. Through a multitude of measurements, the autoprobe enables
the researchers to quickly characterize whether the nanotubes
are semiconducting or metallic.
PEG SKORPINSKI PHOTO |
After spending two years tweaking the combination of materials
to yield strong connections between the various components, the
team produced a one-square-centimeter chip containing thousands
of carbon nanotube transistors accessible via a network of silicon
transistors. “The circuit is interconnected in such a way
that only 22 control signals are needed in testing more than 2,000
nanotubes,” says Tseng. “The key is that this can
all be done by a machine and a computer.”
The researchers probe the nanotubes using commercially available
semiconductor test systems and automatic wafer-probing systems
in their laboratory. Through a multitude of measurements, the
nanotubes can quickly be characterized based on their conductivity.
“We can now grow a lot of the tubes, characterize them quickly,
try something a little different in the growth process, and then
characterize them again,” Bokor says.
The researchers are quick to point out that the test chip is only
the first application of their success integrating silicon circuitry
with carbon nanotube transistors. For example, carbon nanotubes
can be coated with a specific material so that particles of various
environmental agents or chemicals—pathogens in the air,
for example—stick to the outside of the tubes like barnacles
on a ship. As the molecules bind to the carbon nanotubes, the
electrical properties of the tubes change.
“They become selective chemical sensors,” Bokor says.
“And once you integrate them with electronics as we have,
you can imagine a chip that could signal the presence of thousands
of different chemicals.”
Beyond handheld hypersensitive chemical sensors, integrating
silicon technology with nanotube devices could also lead to extremely
dense memory arrays capable of storing tens of thousands times
more information in the same space occupied by a standard memory
chip. Integrated electronics like those in the RANT chip could
read and write data to a compact grid of nanotubes, with each
tube storing a single bit of data.
“Specialized applications like sensors and memory . . .
seem like a sensible approach to introduce carbon nanotube electronics
into a commercial environment,” Bokor says.
With these applications in the back of the researchers’
minds, their next step is to hone their fabrication methods. Already
they’re studying other possible interconnect materials that
aren’t as finicky as molybdenum when it comes to mass production.
They hope future generations of their test chip will lead to the
development of nanotube transistors that can be fabricated in
bulk and offer improved performance over their silicon counterparts.
 |
In
a magnified view of carbon nanotube grown on silicon MOS circuitry,
the bright area on the upper right-hand side is the catalyst
island upon which the nanotube was grown.
IMAGE COURTESY OF ALI JAVEY, STANFORD UNIVERSITY |
“One of the key topics going forward will be in the area
of technology benchmarking. By this I mean comparing new technology
such as carbon nanotubes and silicon with the incumbent silicon
technology itself,” says H.S. Philip Wong, senior manager
of the Exploratory Devices and Integration Technology Department
at the IBM Thomas J. Watson Research Center. “If there is
an advantage, then there is a market. More work such as [the RANT
project] needs to be done.”
The interdisciplinary research is funded by the Massachusetts
Institute of Technology-based MARCO Materials, the Structures
and Devices Focus Center, and the Defense Advanced Research Projects
Agency (DARPA) Microsystems Technology Office (MTO). The project
is part of a larger effort within the UC Berkeley-based Center
for Information Technology Research in the Interest of Society
(CITRIS) to bring an engineering perspective to nanoscience. Indeed,
the RANT research will continue in a new state-of-the-art microfabrication
laboratory slated for construction within a new CITRIS building
planned for campus.
“The days when large companies could have experimental laboratories
for long-range integrated circuit research are gone,” Bokor
says. “We’d like the new microlab to be able to handle
those kinds of efforts.” The new microlab, which will include
an 18,000-foot, two-story clean room, will complement the Molecular
Foundry Nanostructures User Laboratory at Lawrence Berkeley National
Laboratory. Once completed, the new microlab will be a hub for
Berkeley’s groundbreaking efforts in nanoscience and nanoengineering.
“Nanoscience is powerful because it gives us control over
the most fundamental physical properties of matter,” Tseng
says. “Having that control enables us, as engineers, to
develop extremely interesting new devices with unique capabilities.”
David Pescovitz
writes Lab Notes, the College of Engineering's online
research digest, and contributes to Popular Science, Small
Times, and Business 2.0. His writing on science
and technology has been featured in Wired, Scientific American,
IEEE Spectrum, and the New York Times.
|
